1.1. The necessity to study CL prevention behaviors in female students

The outcomes of the appearance of scar production on visible parts of the body are more important in women than that those in men [10]. Unsightly wounds and scars on the face of young women are not only psychologically and socially offending for them [11], but also cause stigma and, in particular, reduce their chances of getting a partner for marriage [3, 12]; on a number of occasions, this stigma leads to their isolation [10]. As relevant research literature indicates, adolescent girls suffer more compared to others in terms of psychological adverse effects and experience of stigma caused by CL [10, 11]. On the other hand, most adolescent girls are students. Therefore, paying attention to CL prevention behaviors in this group can be a priority. In this regard, a reliable and valid tool is needed for the assessment and evaluation of educational programs related to preventive behaviors.

1.2. Theoretical framework selection

With the purpose of identifying CL prevention behaviors, some researchers have used the Health Belief Model (HBM) and individual predisposing factors, such as perceived susceptibility, perceived severity, and self-efficacy, to study some parameters, such as attitude and intention used the BASNEF model [13, 14]. Each of these frameworks includes only a limited number of factors and variables; they do not consider the impact of different types of educational messages. However, in addition to the possibility of measuring the above-mentioned variables, the Extended Parallel Process Model (EPPM) provides the possibility to measure response efficacy and impact of different types of educational messages on the form of audience reactions, including "danger control" and "fear control" [15]. According to the EPPM, it is more likely that individuals efficiently apply health-associated behaviors for danger control providing that they believe their health is at serious risk and trust the efficacy of coping strategies and their self-efficacy to adapt those strategies and preventative behaviors; under this condition, there would not be fear control process stages [15–17]. Although EPPM appears as an appropriate model for the study of individual factors of behavior, the role of enabler factors and social environment should also be considered in identifying CL prevention behaviors [18]. According to studies performed in the form of PRECEDE and PEN-3 models, all the three factors of perceptions, enablers, and nurtures have an important role in the continuation of behaviors, including CL prevention behaviors [5, 19]. The PEN-3 Model consists of three domains: (1) cultural identity, (2) relationships and expectations, and (3) cultural empowerment. Each domain includes three factors that form the acronym PEN, namely the following: person, extended family, neighborhood (cultural identity domain); perceptions, enablers, and nurturers (relationship and expectation domain); positive, existential, and negative (cultural empowerment domain) [19]. This model initially identifies positive, neutral, and negative entities for observing behavior in society, which could be conducive to the prevention of indigenous diseases [19], In addition to perceptions (perceived susceptibility, perceived severity, and self-efficacy,..), this cultural model considers incentives or “nurtures” and also “enablers”, for preventive behaviors [20]. According to these explanations, developing a questionnaire based on the integration of PEN-3 Model and EPPM (Fig 1) would be useful for examination of a wider range of social and individual factors relevant to CL prevention behaviors and evaluation of the impact of interventions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The pattern of PEN-3 and EPPM models integration used in the present paper.

https://doi.org/10.1371/journal.pone.0273400.g001

Therefore, based on the above points, the present study aimed to develop and psychometrically assess Cutaneous Leishmaniasis prevention behaviors questionnaire with application of integration of cultural PEN-3 Model and Extended Parallel Process Model in adolescent female students in endemic areas.

2.1. Initial questionnaire item generation

The item generation was performed via a survey in the social and cultural conditions of the community and by reviewing the results CL studies around the world [21] in addition to the related questionnaires [9, 14, 18, 22]. Afterwards, the pool of questions with 110 items was made in Persian Version, which is the native language of the target group. The itemization and dimension determination of the questionnaire was performed based on the integrated PEN-3 Model and EPPM according to the guidelines and scientific texts related to these models [16, 19, 20, 23, 24].

2.2 Content validity assessment

The content validity was evaluated qualitatively by eight experts (five health education specialists and three healthcare workers) after a careful review of the items with an emphasis on grammar, wording, correct item placement, and scoring. Their written corrective views were applied in the questionnaire.

For the content validity which was performed quantitatively, 14 experts (10 health education specialists and four healthcare workers) were asked to choose "Essential", "Useful but not essential", and "No Necessary" for each item based on the following formula: CVR = [ne − (N/2)] / (N/2). Following the content validity ratio (CVR) calculation, the items with a CVR of higher than 0.51 were accepted according to the Lawshe formula [25]. The same 14 experts were then asked to review and specify the following features for each item in a four-point Likert scale: 1. relevance (1 = irrelevant, 2 = relatively relevant, 3 = relevant, 4 = quite relevant), 2. simplicity (1 = not simple, 2 = relatively simple, 3 = simple, 4 = quite simple), and 3. clarity (1 = not clear, 2 = relatively clear, 3 = clear, 4 = quite clear).

Subsequently, content validity index (CVI) was calculated based on the following formula: CVI = the number of experts who scored the items 3 or 4 / the total number of the experts. The CVI value above 0.79 was considered acceptable for the items [25]. The content validity control with qualitative and quantitative methods made it possible to eliminate unnecessary and irrelevant questions. The items which were considered ambiguous by the experts were reformulated based on the suggested refinements, and questionnaire entered the next stage with 82 items.

2.3. Face validity assessment

The face validity was evaluated qualitatively by the eight above-mentioned experts and the participation of five students via telephone (face-to-face visit was impossible due to the outbreak of COVID-19). After providing the participants with the necessary explanations about the research, their comments on the comprehensibility of each questionnaire item were recorded and the necessary corrections were made. A revised version was then presented to five other students who were asked to express their opinion on the clarity of the items. The results revealed that no more corrections were needed for the items [26].

Thereafter, 10 students were asked to comment on the importance level of each item to quantify the face validity of the questionnaire on a five-point Likert scale from “not important at all” (score = 1) to “very important” (score = 5). Afterwards, the impact score of each item was determined by calculating the impact factor in relative frequency; all the items were deemed appropriately and kept in the questionnaire since they got a score greater than 1.5 [25].

2.4. Construct validity assessment

2.5. Internal and external reliability assessment

Cronbach’s alpha was calculated to determine internal reliability while ICC (intraclass correlation coefficient) was used for the calculation of external reliability.

2.6 Data collection for factor analysis

Since the schools were closed due to the COVID-19 outbreak, the questionnaire was applied in https://porsline.ir to be completed by the students. For this purpose, the short link (yun.ir / il8i51) of the questionnaire was made from its original link (https://survey.porsline.ir/s/ENrbOQE) and it was sent to the high school first-grade female students through virtual networks. In the first part of the questionnaire and after introducing the researcher and other colleagues, some explanations were provided about the purpose of the study, how to participate in the study, and confidentiality of the information; the participants were also informed that they could leave the study whenever they wanted to and that their participation was optional. In the process of answering the questionnaire, they would move throughout the items provided that the consent box was marked.

Two months later, the external reliability of the instrument was calculated via test-retest method. To this end, the same questionnaire was prepared in https://porsline.ir and its link (yun.ir/zzqodb) was sent via SMS to the parents of the students who completed the first series of the questionnaire and voluntarily entered the phone number of one of their parents. They were asked to complete the second questionnaire and the link was deactivated after 31 individuals completed the questionnaire.

2.7. Statistical analysis method

SPSS ver.24 was utilized for statistical analysis. For this purpose, the adequacy of the sample size and the correlation between the extracted factors were evaluated with the Kaiser-Meyer-Olkin (KMO) and Bartlett’s tests. In addition, interpretable factors were identified using exploratory factor analysis (EFA) with varimax rotation. The factors were then identified by the cut-off point of 1.5 for Eigenvalue and cut-off point factor loads were considered 0.3. The Cronbach’s alpha values between 0.70 and 0.95 were satisfactory regarding the external reliability while the internal reliability was determined using intraclass correlation coefficient (ICC) with a two-way random method and a confidence interval of 0.95; a coefficient above 0.70 was considered to have an excellent stability [25].

3.1. Results of the preliminary

The initial version of the questionnaire had 110 items; the perception section on the Likert scale comprised strongly agree, agree, disagree, and strongly disagree and included 20 questions on perceived susceptibility; there were also nine questions on perceived severity, 10 on self-efficacy, and nine on response efficacy. Six questions of nurture, 13 of enablers, and eight of fear were also on the five-point Likert scale (zero, low, medium, high, very high). Fear control questions were in the four-point Likert ranging from strongly agree to strongly disagree. They included six questions about defensive avoidance, three for message minimization, and five for perceived message manipulation. Danger control questions included five questions concerning attitude on a four-point Likert scale (from strongly agree to strongly disagree), four questions about the intention on a five-point Likert scale (from this month, from next month, after next six months, from next year, never), and 12 questions regarding behavior on a four-point Likert scale (never, sometimes, most of the time, always).

3.2. Results of content and face validity assessment

The items with a content validity index (CVI) and content validity ratio (CVR) lower than 0.79 and 0.51, respectively, were removed and ultimately, the content and face validity of the questionnaire was confirmed with 82 items.

3.3. Results of construct validity assessment (exploratory factor analysis)

The result of the KMO test showed that the data volume for EFA was desirable. The value of this index was equal to 0.896. Bartlett’s test was significant (P <0.001, df: 2485) and suggested a sufficient correlation between the variables. At this stage, 11 items were removed due to low factor load and 71 items were entered into the final analysis. Employing Scree diagram and based on the number of common factors and special values above one, 12 interpretable factors were obtained. The factor load of the items was in the range of 0.422 to 0.889 and the total variance of the 12-factor model was 60.50% (Table 1). The factors were named according to the concepts obtained from the loaded items in each factor and also by reviewing the relevant research literature as follows:

• The first factor, "Perceived efficacy", with the variance of 20.949, consisted of 15 items and expressed the perception of the individuals concerning the effectiveness of the recommended behaviors in preventing CL, as well as their understanding of their ability to perform the aforementioned behaviors.
• The second factor, "Behavior", with the variance of 9.107, consisted of nine items and measured the preventive behaviors of CL.
• The third factor was "Message Minimization and Perceived Manipulation" with the variance of 6.006. It comprised eight items and measured the person’s perception of the value and accuracy of CL-related messages. As shown in Table 1, one of the items in the fourth factor (I do not care about the written and visual material about the risks of the seeker) was also included in this group since it is conceptually more similar to this category.
• The fourth factor, "Defensive Avoidance of the Message", with the variance of 4.640, consisted of five items. It indicated how much a student blocks the path for receiving messages about CL and avoids receiving such messages.
• The fifth factor was "Nurture", including five items with the variance of 3.381. It measured the role of other people in strengthening and encouraging a person in performing CL prevention behaviors.
• The sixth factor, "Perceived Severity", with the variance of 3.107, consisted of seven items related to students’ perception of the severity of the disease and its complications.
• The seventh factor, "Fear", with the variance of 2.636, comprised four items and measured anxiety, anger, and worrisome caused by fear of CL messages and their complications.
• "Intention" was the eighth factor with the variance of 2.576. It included four items that measured the intention to perform preventive behaviors of CL.
• The ninth factor, "Perceived Susceptibility in case of Insufficient Personal Protection" with the variance of 2.325, included four items that measured the students’ perception of the risk of CL in case of insufficient personal protection.
• The 10th factor was “Individual Enablers” with the variance of 2.123. It had four items that measured the students’ access to facilities and their personal skills to prevent CL.
• The 11th factor, "Environmental Enablers", with the variance of 1.862, contained four items related to environmental facilities whose availability enabled the individual to prevent CL. These include installing an efficient net into the doors, windows, and air conditioner vents, as well as a timely collection of construction waste and garbage by the related organizations.
• The 12th and last factor was "Perception of the Possibility of Bites at Any Height" with the variance of 1.787. It comprised two items about the students’ knowledge about the fact that living on the upper floors or resting on the bed will also put them at risk of sandfly bites.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Items of the questionnaire of the CL preventive behaviors and factors related to the integration of PEN-3 and EPPM with mean, Std. Deviation, CVR, and CVI.

https://doi.org/10.1371/journal.pone.0273400.t001

Factor analysis was performed to explore the main factors of the questionnaire. The factor load of each question as well as internal and external reliability factors, along with the Eigenvalue, Variancepercentage, Variance Cumulativepercentage are reported in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Factor analysis results for the exploration of the main factors of the CL-preventive behaviors questionnaire and factors related to the integration of PEN-3 Model and EPPM, Cronbach’s α coefficient, and ICC.

https://doi.org/10.1371/journal.pone.0273400.t002

3.3. Results of internal and external reliability assessment

The internal reliability of the instrument was calculated after factor analysis through Cronbach’s alpha for the whole questionnaire and for each factor. The alpha coefficient for the 12 above-mentioned factors were 0.919, 0.888, 0.898, 0.852, 0.848, 0.813, 0.903, 0.913, 0.710, 0.653, 0.694, and 0.646, respectively; meanwhile, it was 0.877 for the whole instrument. Therefore, the internal reliability of the instrument was confirmed (Table 2).

The external reliability of the instrument was determined via ICC coefficient, which were 0.792, 0.767, 0.764, 0.712, 0.729, 0.740, 0.741, 0.710, 0.663, 0.817, 0.564, and 0.553 for the first to 12th factor, respectively. It was calculated as 0.833 for the whole instrument; thus, the external reliability of the instrument was confirmed (Table2).

4.1. Extracted components based on EPPM

4.2. Extracted components based on PEN-3 model

The dimensions of the PEN-3 model in this questionnaire included Perceptions, Enablers, and Nurture. The Perceptions were consistent with Perceived Efficacy, Perceived Susceptibility, and Perceived Severity in the EPPM, as described in the previous sections.

4.3. Study limitations

One of the difficulties faced conducting the present study was the COVID-19 pandemic that prevented face-to-face interviews for face validity; it was performed via phone calls. Furthermore, we had to complete the questionnaires online instead of doing so in person. Failure to perform confirmatory factor analysis was another limitation herein. It could be suggested that confirmatory factor analysis be performed in future studies.